The manuscript documented a regional modeling effort using the ISAM to quantify carbon fluxes from the spring wheat agroecosystems in India. Overall, the manuscript is well organized and the topic is of interest to the community. However, the following major concerns should be addressed before the manuscript can be considered for publication.

First, there is no validation of spring wheat yield in the manuscript, which is a major carbon flux out of the agroecosystem. I strongly suggest the authors to add the validation of yield at both site and regional scale to demonstrate that the yield can capture the variation of this important carbon flux. Related to this suggestion, please also add how the model simulates yield formation processes in the method section.

Second, for the long-term simulation of agroecosystem, crop rotation is a critical factor as it will affect the soil biogeochemical cycling and thus the long-term soil fertility. However, this part is largely unaddressed in the current manuscript. Besides spring wheat, what other crops are planted in the cropping systems in reality and how was that handled in the ISAM modeling efforts? Without simulating the typical crop rotation, I don’t think the carbon fluxes can be reliably simulated by the model.

Third, changes of crop cultivars and management practices (as well as their spatial variations) are not well considered in the manuscript. For long-term simulation, these factors are critical aspects that cannot be neglected, especially when the focus is related to carbon.

Fourth, the authors are using the dynamic planting date predicted by the model, however, the authors are not evaluating whether the simulated sowing date is reflecting the reality. The authors should have access to several crop calendars and also have good knowledge of the local farming seasonality. I would suggest the authors to validate the predicted sowing date as it is such a critical factor affecting the spatial pattern of carbon fluxes shown in Fig. 3. Otherwise, I can not have more confidence in the spatial patterns of carbon fluxes, which are not well interpreted by the authors.

Finally, before showing the spatial pattern and temporal trends of carbon fluxes, there are so many other intermediate variables which should be checked, such as leaf area index, biomass, and crop yield.

Other comments:

Fig. 1. Why did the authors only show monthly data here? Daily time series of carbon fluxes can also be added here.

L124 and L134: what’s the criteria of steady state of soil parameters? The authors should demonstrate that by plotting the data.

Fig. 2: what is leading to the systematic bias here?

Fig. 4: Is the Ra here too low? Rule of thumb is that NPP=0.5GPP, which indicates that Ra~0.5GPP.

Authors of this manuscript use ISAM model calibrated over a wheat site to explore the carbon flux change over Indian spring wheat region since 1980. They further performed factorial simulations to attribute the carbon flux change. Overall, the manuscript addresses an interesting topic, but the quality of the study and the presentation need to be improved before it could be acceptable.

Some critical details of the model and modelling experiments were missing from the manuscript. For example, the author stated that ISAM_dyn_whe with dynamic phenology, carbon allocation and vegetation phenology. However, how these modules were formulated remain unknown.

The authors simulate three decades’ change of cropland carbon flux, but how change in crop varieties and management practices was accounted remain unknown. If these changes were not accounted, the simulated change in the carbon flux could be far away from the reality.

Carbon fluxes over croplands heavily depend on phenology and managements. These conditions could vary largely from year to year. While the authors recognize the importance in accounting them, in calibrating and validating their model, the phenology and flux data driving the model come from different years. This should introduce biases/uncertainties.

While calibration of the crop model in a site with good observation is helpful for robustness of model simulation results. However, using the model calibrated on one site to represent the entire Indian spring wheat region is far from giving readers good confidence. There are many satellite observations and statistics available to test model performance (e.g. LAI, FPAR and yield), which should be used to validate the model in regional applications.

Attribution of carbon flux change to climate variations at regional scale have strong spatial heterogeneity. A simple bar figure is not very informative, in particular for changes in climatic variable.